A Putative Mechanism for Magnetoreception by Electromagnetic Induction in the Pigeon Inner Ear

Report

Graphical Abstract Authors Simon Nimpf, Gregory Charles Nordmann, Daniel Kagerbauer, ..., Martin Colombini, Matthew J. Mason, David Anthony Keays

Correspondence [email protected]

In Brief Nimpf, Nordmann et al. conﬁrm that magnetic stimuli result in neuronal activity in the vestibular nuclei of pigeons. Hypothesizing that this is attributable to electromagnetic induction within semicircular canals of the inner ear, they demonstrate the presence of known electrosensory molecules in vestibular hair cells.

Highlights d Magnetic stimuli activate neurons in the caudal vestibular nuclei d Magnetic stimuli induce a voltage in a model of a semicircular canal d Electroreceptive molecules are expressed in vestibular hair cells d We postulate that pigeons detect magnetic ﬁelds by electromagnetic induction

Nimpf et al., 2019, Current Biology 29, 4052–4059 December 2, 2019 ª 2019 The Author(s). Published by Elsevier Ltd. https://doi.org/10.1016/j.cub.2019.09.048 Current Biology Report

A Putative Mechanism for Magnetoreception by Electromagnetic Induction in the Pigeon Inner Ear

Simon Nimpf,1,5 Gregory Charles Nordmann,1,5 Daniel Kagerbauer,2 Erich Pascal Malkemper,1 Lukas Landler,1 Artemis Papadaki-Anastasopoulou,1 Lyubov Ushakova,1 Andrea Wenninger-Weinzierl,1 Maria Novatchkova,1 Peter Vincent,3 Thomas Lendl,1 Martin Colombini,1 Matthew J. Mason,4 and David Anthony Keays1,6,* 1Research Institute of Molecular Pathology (IMP), Vienna Biocenter (VBC), Campus-Vienna-Biocenter 1, 1030 Vienna, Austria 2Atominstitut, TU Wien, Stadionallee 2, 1020 Vienna, Austria 3Department of Neurology, Yale University School of Medicine, 333 Cedar Street, New Haven, CT 06520, USA 4Department of Physiology, Development and Neuroscience, University of Cambridge, Downing Street, Cambridge CB2 3EG, UK 5These authors contributed equally 6Lead Contact *Correspondence: [email protected] https://doi.org/10.1016/j.cub.2019.09.048

SUMMARY activation in the vestibular nuclei of pigeons [5]. To perform this experiment within the laboratory environment, we built a A diverse array of vertebrate species employs the room constructed of mu metal surrounded by an aluminum Earth’s magnetic field to assist navigation. Despite Faraday cage to shield against static and oscillating magnetic compelling behavioral evidence that a magnetic fields (Figure 1A). This setup allowed us to perform experiments sense exists, the location of the primary sensory cells in a controlled, magnetically clean environment [7]. Magnetic and the underlying molecular mechanisms remain un- fields were generated using a double-wrapped, custom-built known [1]. To date, most research has focused on a 3D Helmholtz coil system situated in the center of the shielded room (Figure 1B). To reduce movement during the experiments, light-dependent radical-pair-based concept and a birds were head fixed using a surgically implanted plastic head system that is proposed to rely on biogenic magnetite post, and the body was immobilized using a 3D printed harness (Fe3O4)[2, 3]. Here, we explore an overlooked hypoth- (Figure 1C). We applied the same stimulus as Wu and Dickman, esis that predicts that animals detect magnetic fields exposing adult pigeons to a 150-mT, rotating magnetic field (n = by electromagnetic induction within the semicircular 22) or to a zero magnetic field (n = 23) for 72 min (Figures 1D–1F) canals of the inner ear [4]. Employing an assay that re- [5]. We performed this experiment both in darkness (n = 30) and lies on the neuronal activity marker C-FOS, we under broad-spectrum white light (n = 15). Birds were then confirm that magnetic exposure results in activation perfused, the brains were sliced, and matched sections contain- of the caudal vestibular nuclei in pigeons that is inde- ing the vestibular nuclei (3 sections per bird) were stained with pendent of light [5]. We show experimentally and by sera against the neuronal activity marker C-FOS. To minimize physical calculations that magnetic stimulation can variation, all staining was performed simultaneously, all slides were scanned with the same exposure settings, and C-FOS- induce electric fields in the pigeon semicircular canals positive neurons were counted using a machine-learning-based that are within the physiological range of known elec- algorithm. Using established anatomical coordinates, we troreceptive systems. Drawing on this finding, we segmented the medial vestibular nuclei (VeM) and compared report the presence of a splice isoform of a voltage- the density of C-FOS-positive cells of the experimental groups gated calcium channel (CaV1.3) in the pigeon inner [8]. We observed an increase in the density of C-FOS-positive ear that has been shown to mediate electroreception cells in both the light and dark when exposing birds to magnetic in skates and sharks [6]. We propose that pigeons fields. In the light, the average density in controls was 35.32 ± detect magnetic fields by electromagnetic induction 10.20 cells/mm2 (n = 8; mean ± SD) whereas with the magnetic 2 within the semicircular canals that is dependent on treatment it was 44.38 ± 5.72 cells/mm (n = 7; mean ± SD). In the dark, the average density for control birds was 34.70 ± the presence of apically located voltage-gated cation 2 channels in a population of electrosensory hair cells. 10.82 cells/mm (n = 15; mean ± SD) whereas with the magnetic treatment it was 50.52 ± 27.34 cells/mm2 (n = 15; mean ± SD). An application of a two-way ANOVA revealed a significant RESULTS effect of the magnetic treatment but no interaction between magnetic treatment and lighting conditions (two-way ANOVA; Magnetically Induced Activation in the Pigeon magnetic: p = 0.0176, F = 6.125; magnetic by light: p = 0.5675, Vestibular Brainstem F = 0.332) (Figures 2A and 2B; Table S1). To explore this in We set out to replicate a previous study conducted by Wu and more detail, we employed our spot-detection algorithm coupled Dickman, who reported that magnetic stimuli induce neuronal to an elastic registration to generate heatmaps showing regional

4052 Current Biology 29, 4052–4059, December 2, 2019 ª 2019 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Figure 1. Experimental Setup for the Magnetic Activation Study (A) The magnetic activation experiments were performed inside a magnetically shielded room consisting of a Faraday cage to shield from broadband anthro- pogenic magnetic noise and a mu-metal cage to shield from the geomagnetic field. (B and C) Pigeons were placed in the center of a double-wrapped 3D Helmholtz coil system (B) located in the magnetically shielded room and immobilized by head fixation and a molded body harness (C). (D–F) Schematic illustration of the magnetic stimulus used in this study. The magnetic field vector was rotated 360 within a plane by 6 steps every 2 s (D). After each rotation the plane was shifted by 15 (E) and the procedure was repeated until a full rotation around each axis was achieved (12 planes per x, y, and z axis, respectively) (F). differences in the density of C-FOS-positive cells. This revealed nanovoltmeter (forming a loop with a diameter of 21 cm), placed an enrichment of activated neurons in the dorsomedial part of in the center of our magnetic coil system, and exposed to the the VeM in animals exposed to magnetic stimuli (Figure 2C). Pre- same rotating magnetic stimulus applied to our birds (i.e., vious tracing experiments have shown that the dorsomedial VeM 150 mT rotating 360 with 6-step changes every 2 s). We is innervated by projections from both the semicircular canals observed discrete voltage spikes when the magnetic field stim- and the otolith organs located in the inner ear [9]. ulus was presented and with each stepwise change (Figure 3B). In accordance with Faraday’s law of electromagnetic induction, A Model for Electromagnetic Induction in the Pigeon we observed the maximum induced voltage (15.6 mV) when the Inner Ear magnetic field vector was perpendicular (90) to the plane of As we did not observe an interaction between the presence of the canal and lowest (1.6 mV) when the vector was parallel to light and magnetically induced neuronal activation, our results the plane of the canal (0)(Figure 3B). In contrast, when present- are consistent with a magnetic sensory system based on either ing the control stimulus (with the current running antiparallel magnetite or electromagnetic induction [10]. We have previously through the double-wrapped coils), we did not observe these reported the discovery of an iron-rich organelle in both vestibular characteristic voltage spikes (Figure 3C). Nor did we observe and cochlear hair cells that is associated with vesicular struc- voltage spikes when removing the model of the semicircular ca- tures, but because it is primarily composed of ferrihydrite it lacks nal from the circuit, demonstrating that induction does not occur the magnetic properties to function as the hypothesized torque- in the connecting shielded wires (Figure 3D). based magnetoreceptor [11–13]. Moreover, a systematic screen Next, we measured the dimensions of the semicircular canals for magnetite in the pigeon lagena using synchrotron-based in pigeons drawing on previously generated computed tomogra- X-ray fluorescence microscopy and electron microscopy has phy (CT) scans (n = 3 birds) [18](Figure S1A). This revealed that failed to identify extra- or intracellular magnetite crystals [14]. In the mean loop diameter of the posterior canal was 4.20 ± light of these findings, we focused on electromagnetic induction 0.01 mm, the anterior canal was 6.37 ± 0.05 mm, and the lateral [15]. First proposed by Camille Viguier in 1882, this hypothesis canal was 5.18 ± 0.15 mm (mean ± SD). Drawing on these mea- predicts that as a terrestrial animal moves through the Earth’s surements and the voltage induced in our 21-cm-diameter static magnetic field a voltage is induced within the conductive model, we were able to estimate the electric field generated endolymph of the semicircular canals [4, 16]. within a pigeon’s semicircular canal on presentation of the To test the viability of this hypothesis, we built a simple model Wu and Dickman stimulus. We applied the following equation: of a pigeon semicircular canal by filling a plastic tube with artifi- E = ðu =2prÞ, where E is the electric field, r is the mean radius cial pigeon endolymph (see STAR Methods; Figure 3A) [17]. The of the semicircular canal loop, and u is the measured induced tubing was closed on both sides with electrodes connected to a voltage. This revealed a maximal electric field in the anterior

Current Biology 29, 4052–4059, December 2, 2019 4053 Figure 2. Magnetically Induced Neuronal Activation in the Pigeon Vestibular Nuclei (A) Representative coronal section stained with sera that bind to the neuronal activity marker C-FOS. Dark-stained nuclei are visible in the inset. Adjacent schematic representation of the pigeon brainstem shows the medial and de- scending vestibular nuclei (VeM and VeD) in gray [8]. (B) Quantification of C-FOS-positive cells per mm2 in VeM in birds that were exposed to a rotating magnetic field (MF; n = 22) and to a zero magnetic field control (ZMF; n = 23) (two-way ANOVA; p = 0.0176, F = 6.125). Filled and open circles represent animals that were tested in the dark (n = 15 for ZMF andMF) and in light (n = 8 for ZMF and n = 7 for MF), respectively. (C) Heatmap showing the regional difference in C-FOS cell density in the vestibular nuclei when comparing the MF and ZMF groups. Magnetically responsive neurons were enriched in the dorsomedial region of the VeM. Data are presented as mean ± SD. *p < 0.05. Scale bars represent 500 mm(A)and50mm (A, inset). VeM, vestibular nucleus medialis; VeD, vestibular nucleus descendens; TTD, nucleus et tractus descendens nervi trigemini; NTS, nucleustractus solitarius; nIX-X, nucleus nervi glossopharyngei et nucleus motorius dorsalis nervi vagi. See also Table S1. canal of 7.2 nV/cm, 5.8 nV/cm for the lateral canal, and 4.7 nV/cm electrosensory organs found in cartilaginous fish that consist for the posterior canal when applying a rotating 150-mT stimulus. of specialized sensory cells located at the base of gel-filled ca- We repeated the aforementioned experiment, applying a 50-mT nals [22]. Recent work in little skate (Leucoraja erinacea) has stimulus, again rotating 360 in 2 min with 6 -step changes every shown that the voltage-sensitive calcium channel CaV1.3/CAC- 2 s. This revealed a maximal electric field in the anterior canal of NA1D and the large-conductance calcium-activated potassium 2 nV/cm, 1.6 nV/cm for the lateral canal, and 1.3 nV/cm for the channel BK/KCNMA1 are enriched in these cells and facilitate posterior canal (Figures S1B–S1D). electrosensation [23]. Given the ontogenetic proximity of electro- In our experiment, we applied a changing magnetic stimulus to sensory, auditory, and vestibular hair cells, we asked whether a bird that is head fixed; however, in the natural environment, we these electrosensory molecules are present in the pigeon semi- imagine that pigeons employ head scanning to alter the orienta- circular canals [24]. To function as electroreceptors, we would tion of their semicircular canals with respect to the magnetic vec- expect these channels to be apically located in hair cells where tor. To assess whether this natural behavior would be sufficient they would be exposed to the endolymph (Figures 4A–4C). To to induce electric fields that are physiologically relevant, we esti- test this prediction, we performed fluorescence immunohisto- mated the electric field generated using the following equation, chemistry on pigeon ampullary hair cells (n = 3 birds). Staining which is derived from Maxwell’s third law: En = B0,pf,r, where with sera against the BK channel revealed an enrichment of api- r is the radius of the semicircular canals, B0 is the Earth’s field, cal staining in hair cells that are positive for the marker otoferlin and f is the frequency of head scanning [19]. It has recently (Figures 4D, 4F, 4G, and S2D–S2I) [25, 26]. Consistent with pre- been shown that pigeons undertake head scanning during flight vious studies, we observed goblet-shaped CaV1.3 staining at that exceeds 700/s [20]. Based on this frequency and the radii of hair cell ribbon synapses [27] and a punctate enrichment at the the semicircular canals reported here, we estimate that natural apical membrane that has not previously been reported (Figures head movement will generate electric fields in the range of 7.9– 4E and 4G). High-resolution confocal imaging further revealed

9.6 nV/cm in an Earth-strength field (50 mT). Because previous that these CaV1.3-rich plaques are most pronounced at the studies have shown that electrosensitive animals can detect base of the kinocilium, and that CaV1.3 is localized in the kinoci- fields as small as 5 nV/cm [21, 22], we conclude that the Wu lium itself (Figures S2A–S2C). and Dickman stimulus and natural head-scanning behavior In sharks and skates, CaV1.3 is characterized by a low could induce voltages within the semicircular canals that are threshold of activation attributable to a 10-amino acid lysine- within the detectable range of known electrosensory systems. rich insertion located in the intracellular loop between IVS2 and IVS3 of the alpha subunit (Figure 4H). It has been shown that Electroreceptive Molecules Are Expressed in the Pigeon mutating the charged lysine residues in this insertion to neutral Inner Ear glutamine residues results in a channel with a higher threshold If magnetoreception in pigeons relies on the conversion of for activation [6, 23]. Interestingly, Hudspeth and colleagues magnetic fields into an electric signal, the magnetosensory have reported a similar insertion in CaV1.3 that is expressed in apparatus might resemble known electroreceptive epithelia on the hair cells of chickens [28]. To ascertain whether this charged a cellular and molecular level. Ampullae of Lorenzini are insertion is expressed in pigeons, we drew on the available

4054 Current Biology 29, 4052–4059, December 2, 2019 Figure 3. Modeling of Electromagnetic In- duction in the Pigeon Inner Ear (A) A model of a pigeon semicircular canal (21 cm in diameter) containing artificial endolymph was exposed to a 150 mT rotating magnetic field. The stimulus consisted of 6 shifts every 2 s, completing one rotation in 2 min around the vertical axis. The induced voltage was measured using a nanovoltmeter. (B) We observed voltage spikes correlating with changes in the magnetic field. The peak amplitude was highest (15.6 mV) when the stimulus was directed 90 to the plane of the canal and lowest (1.6 mV) when the vector was parallel to the plane of the canal. (C) These voltage spikes were absent when the control stimulus was presented with currents running antiparallel through the double-wrapped coils. (D) Voltage spikes were likewise absent when the 150 mT rotating magnetic field was exposed to a circuit that did not include the loop containing the artificial endolymph. This shows that induction does not occur in the connecting wires. See also Figure S1.

genomic resources, determined the gene structure, and de- Phylogenetic Analysis of CaV1.3/CACNA1D and BK/ signed a PCR-based strategy with primers flanking the putative KCNMA1 insertion site (Figures 4H and S3A) [29]. We extracted mRNA and Finally, we explored the phylogenetic distribution of the splice generated cDNA libraries for a broad range of tissues from the isoforms of CACNA1D and KCNMA1 among Animalia. To iden- pigeon including brain, spleen, muscle, retina, basilar papilla, tify CACNA1D homologs containing the long KKER splice vestibular epithelia, skin, and heart (n = 3 birds). Analysis by isoform, we performed a BLASTp search with a 50-amino- gel electrophoresis showed that CaV1.3 is absent in the respira- acid-long segment of pigeon CaV1.3, centered around the splice tory concha, muscle, and liver but is otherwise broadly ex- isoform. We found that this insertion first emerged in Gnathosto- pressed. Strikingly, we only observed PCR products consistent mata and has a limited distribution. It is found in skates, sharks, with a larger splice isoform in the cochlea and vestibular system bats, turtles, rainbow trout, and birds (Figures 4I and S4A). In the (Figure 4I). Cloning of these PCR products revealed an insertion case of the KCNMA1 RA isoform, a BLASTp search revealed that of 10 amino acids rich in lysine residues with notable homology it is present in numerous animals including skates, electric eels, to that reported in sharks and skates (Figure 4J). We refer to numerous fish and many bird species, rodents (including Mus this variant as the KKER splice isoform. musculus), primates (including Homo sapiens), bats, dolphins,

In skates it has been shown that CaV1.3 works in conjunc- and whales (Figures 4L and S4B). We conclude that the RA splice tion with the BK potassium channel, which has a unique isoform of KCNMA1 is widely distributed in vertebrates whereas conductance profile due to the expression of an alternatively the KKER isoform of CACNA1D is often found in phyla that are spliced exon [23]. This variant is distinguished by the presence known to possess an electric or magnetic sense. of an arginine residue at position 340 (R340) and an alanine residue at 347 (A347), which are located intracellularly near DISCUSSION the pore of the channel (Figure 4K). We refer to this variant as the RA splice isoform. To ascertain whether this splice iso- In 1882, Viguier speculated that ‘‘the geomagnetic field deter- form is expressed in pigeons, we first determined the genetic mines, within the endolymph of the canals, induced currents, architecture of pigeon KCNMA1 (Figure S3B). Drawing on this whose intensities vary dependently of both the canals’ positions information, we designed a PCR-based assay that relies on in relation to inclination and declination, and the intensity of the the amplification of exon 9a/b from cDNA libraries, followed magnetic field’’ [4]. In this manuscript, we have explored this hy- bydigestionwiththerestrictionenzymeAluI(Figures 4Land pothesis, one that has largely been ignored by the scientific com- S3B). In the event the RA isoform is amplified (exon 9b), the munity since its proposition. We present data that replicates the amplicon is resistant to AluI digestion, resulting in the pres- work of Wu and Dickman, demonstrating magnetically induced ence of a larger band (239 bp). We found that KCNMA1 is neuronal activation in the vestibular nuclei that is not dependent broadly expressed in the pigeon; however, the RA isoform on light. We show that changing low-intensity magnetic stimuli was enriched in the retina and vestibular system. We conclude (150 mT) can induce electric fields that lie within the window of that the molecular apparatus necessary for the detection of physiological detection and that the molecular machinery neces- small electric fields is present within the pigeon vestibular sary to detect such fields is present in the pigeon inner ear. Our apparatus. data are consistent with a model whereby pigeons detect

Current Biology 29, 4052–4059, December 2, 2019 4055 Figure 4. Electroreceptive Molecules Are Expressed in the Pigeon Inner Ear (A–C) Schematic illustrating induction-based magnetoreception in the pigeon inner ear. (A) A semicircular canal with sensory hair cells located at the base of the gelatinous cupula at the crista ampullaris (ca). (legend continued on next page) 4056 Current Biology 29, 4052–4059, December 2, 2019 magnetic fields by electromagnetic induction within the semicir- gene expression profiles, and share characteristic ribbon synap- cular canals relying on the presence of apically located voltage- ses [31, 32]. It is therefore conceivable that hair cells in Aves have gated calcium channels in a population of electrosensory hair maintained or acquired an electrosensory capacity that is ex- cells. ploited for magnetic detection. We have replicated the study of Wu and Dickman with only mi- A critical issue in considering the validity of the inductive hy- nor changes to their experimental protocol [5]. Specifically, we pothesis is how magnetic information would be distinguished employed double-wrapped coils to control for heat and vibration from vestibular input assuming that ampullary hair cells are when delivering the magnetic stimuli and a plastic head post and both mechanically and electrically sensitive [4, 16]. The anatom- glue in preference to metal screws to head fix the bird and con- ical framework of the semicircular canals provides an elegant ducted the experiment in both darkness and light. We have solution to this problem. Imagine a magnetic vector with an further been able to refine the region activated by magnetic stim- orientation that is perpendicular to the plane of a semicircular ca- uli by elastic registration and mapping of C-FOS-positive cells to nal. Rotation in the plane of the canal leads to inertia-based fluid the dorsomedial part of the VeM. Our results, coupled with the displacement and mechanic stimulation of hair cells but minimal initial study, support the contention that the vestibular system electromagnetic induction. In contrast, rotation perpendicular to is involved in processing magnetic information in the absence the plane does not result in fluid movement but generates elec- of light. Wu and Dickman had argued that the primary sensors tric field changes [16]. In this way, magnetic and vestibular infor- likely reside in the lagena, because extirpation of the cochlear mation originating from the same sensory epithelia could be duct abolished magnetically induced activity in the vestibular distinguished from each other, so long as the animal has a third nuclei. These results, however, are also consistent with magne- reference frame (e.g., visual input). A second important element tosensation by electromagnetic induction, because removal of to this model is the presence of the cupula within the crista the cochlear duct would compromise the integrity of the entire ampullaris. Acting as a physical barrier it enables the separation endolymphatic system and its ionic constituents. of positively charged cations (Na+,K+) as the animal moves its Magnetosensation by induction has only been considered head through the magnetic field [16]. The permeability of this viable in elasmobranch fish, because surface-electrosensitive structure to cations has yet to be determined, but we assume epithelia could detect voltages induced as the animal moves that its gelatinous composition and its positive charge impede through the conductive seawater within the Earth’s static mag- cation flow [33, 34]. netic field [21, 22]. In birds, induction has largely been dismissed An experimental paradigm that permits investigators to distin- due to the high electrical resistivity of the air and the lack of a guish between magnetoreception based on induction and conductive circuit [1]. Our physical modeling and anatomical magnetite is to fix a strong magnet (e.g., 10 mT) to the bird. measurements suggest that the semicircular canals possess Such a field would immobilize any magnetite chains, rendering the requisite dimensions and properties to function as a physio- them unresponsive to the application of an Earth-strength field. logical dynamo. Moreover, the apical location and the expres- In contrast, even a small magnetic vector superimposed onto a sion of splice isoforms of CaV1.3/CACNA1D and BK/KCNMA1 larger fixed field would result in a changing magnetic stimulus indicate that hair cells in pigeons may also function as electrore- and permit magnetoreception by electromagnetic induction. ceptors. It is known that hair cells are closely related to electro- Although this experiment has yet to be performed in a controlled sensory cells on a developmental, cellular, and molecular level laboratory-based assay in pigeons, a number of groups have [24]. The ancestral lateral line system of vertebrates consists of attached magnets to birds to assess their effect on homing both mechanosensory hair cells and electrosensory ampullary and navigation. Keeton glued bar magnets onto the backs of pi- organs, which develop from lateral line embryonic placodes geons, initially reporting that this interfered with their homeward [30]. Both cell types have apical ciliary protrusions and similar orientation on overcast but not sunny days [35], which was

(B) Rotation of the bird’s head perpendicular to the plane of the semicircular canal causes a redistribution of charges across the cupula (cu). (C) Voltage-gated ion channels (VGIC) that are located apically in hair cells respond to this redistribution of charge, by either allowing or restricting cation inﬂux into the hair cell. (D) Histological section of a pigeon crista ampullaris stained with the hair cell (hc) marker otoferlin and nuclear marker DAPI (n = 3 birds).

(E–G) Co-staining with CaV1.3 (E and G) and the large-conductance calcium-activated potassium channel BK (F and G) shows that the voltage-sensitive machinery is expressed in otoferlin-positive hair cells (D and G) and localizes to the apical surface facing the endolymphatic space (es). Note the distinctive apical punctate staining of CaV1.3.

(H) Schematic representation of a pigeon CaV1.3 alpha subunit. The IVS2-S3 domain (shown with an asterisk) is encoded by exon 28 and is subject to alternative splicing.

(I) PCR ampliﬁcation of various cDNA libraries employing primers that ﬂank exon 28. CaV1.3 is broadly expressed in the pigeon, but the larger splice isoform that encompasses exon 28b is only expressed in the cochlea and vestibular epithelia.

(J) Homology alignment of pigeon CaV1.3 shows that this long isoform is characterized by charged residues (KKER), which confer a low activation threshold in skates and sharks. This insertion is absent in zebrafish, Xenopus, mice, and humans. (K) Schematic representation of a pigeon BK channel alpha subunit. The domain proximal to the pore is encoded by exon 9 and highlighted with an asterisk. (L) Gel image of an AluI restriction digest following PCR amplification of cDNA libraries with primers spanning exon 9. Amplicons that include exon 9b contain the RA allele, which is resistant to digestion by AluI, resulting in a single 239-bp band. In contrast, amplicons that encompass exon 9a result in a digestible product that generates two bands (92 and 147 bp). Exon 9a is expressed in all tissues whereas exon 9b (i.e., the RA isoform) is enriched in the vestibular system and the retina. (M) Homology alignment showing that the RA isoform is also found in zebrafish, bats, mice, and humans. Scale bars represent 50 mm (D) and 10 mm (E–G). NC, negative control. See also Figures S2–S4.

Current Biology 29, 4052–4059, December 2, 2019 4057 replicated by Ioale` [36]. However, it should be noted that Keeton B PCR analysis of Cav1.3 (CACNA1D) splice isoforms himself generated contradictory results when performing a larger B PCR analysis of BK (KCNMA1) splice isoforms study between 1971 and 1979 [37]. Several groups have B Phylogenetic analysis of Cav1.3 (CACNA1D) and BK attached magnets to the heads of albatrosses and observed (KCNMA1) splice isoforms no effect on navigation, leading the authors to speculate that d QUANTIFICATION AND STATISTICAL ANALYSIS the birds do not have a magnetite-based magnetosensor, or d DATA AND CODE AVAILABILITY that they do not rely on magnetic cues for orientation [38, 39]. Although these studies have largely ignored the possibility that SUPPLEMENTAL INFORMATION birds might rely on electromagnetic induction as a mechanism, we would urge caution when interpreting their results. There Supplemental Information can be found online at https://doi.org/10.1016/j. are numerous uncontrolled environmental variables, and it is un- cub.2019.09.048. clear whether the magnets were truly fixed in position given they ACKNOWLEDGMENTS were glued to the skin of the birds. Finally, we wish to acknowledge that there are alternative D.A.K. is supported by the European Research Council (ERC; 336725) and the explanations for our results. First, it is conceivable that the FWF (Y726). S.N. is a recipient of a DOC fellowship of the Austrian Academy of magnetically induced activation that we observe in the VeM is Sciences at the Research Institute of Molecular Pathology. We wish to thank a consequence of multimodal sensory integration and not due Boehringer Ingelheim, which funds basic scientific research at the Research to primary sensors located in the inner ear. Second, we have Institute of Molecular Pathology. We are indebted to the excellent support fa- cilities at the IMP and VBCF including histology, bio-optics, bioinformatics, assumed that the light-independent magnetic activation we and the workshop. report excludes a radical-pair-based mechanism; however, it is possible that a chemical-based compass may exist and does AUTHOR CONTRIBUTIONS not depend on light. Moreover, our results do not preclude the existence of a light-based magnetoreceptor in pigeons, because Conceptualization, S.N., G.C.N., and D.A.K.; Investigation, S.N., G.C.N., D.K., they may possess more than one magnetosensory system. E.P.M., L.L., A.P.-A., L.U., A.W.-W., and P.V.; Formal Analysis, S.N., G.C.N., D.K., E.P.M., M.N., and L.L.; Software, T.L.; Resources, M.J.M. and M.C.; Third, the long isoform of CaV1.3 (although critical for electroreception in sharks and skates) may merely tune hair cells to audi- Writing – Original Draft, S.N., G.C.N., and D.A.K.; Writing – Review & Editing, E.P.M., M.J.M., P.V., D.K., and L.L.; Funding Acquisition, D.A.K. and S.N.; Su- tory and vestibular stimuli in birds. Despite these caveats, the pervision, D.A.K. putative mechanism that we present in this paper enables us to make several predictions. Should pigeons rely on CaV1.3 to DECLARATION OF INTERESTS detect magnetic fields by electromagnetic induction, we expect that magnetically induced neuronal activation will be compro- The authors declare no competing interests. mised by (1) pharmacological intervention with CaV1.3 antago- nists such as nifedipine; (2) hair cell ablation with antibiotics; Received: June 26, 2019 Revised: August 28, 2019 and (3) genetic deletion of the long isoform of CaV1.3. In contrast, Accepted: September 19, 2019 neuronal activation will be preserved if a strong dipole magnet is Published: November 14, 2019 fixed on the head of the animal, immobilizing any magnetite par- ticles but still permitting induction from an applied changing REFERENCES stimulus. These predictions will serve as a basis for future experiments and the interrogation of a hypothesis that has been 1. Johnsen, S., and Lohmann, K.J. (2005). The physics and neurobiology of forgotten but not yet falsified. magnetoreception. Nat. Rev. Neurosci. 6, 703–712. 2. Ritz, T., Adem, S., and Schulten, K. (2000). A model for photoreceptor- based magnetoreception in birds. Biophys. J. 78, 707–718. STAR+METHODS 3. Kirschvink, J.L., Walker, M.M., and Diebel, C.E. (2001). Magnetite-based magnetoreception. Curr. Opin. Neurobiol. 11, 462–467. Detailed methods are provided in the online version of this paper 4. Viguier, C. (1882). Le sens de l’orientation et ses organes chez les animaux and include the following: et chez l’homme. Rev. Philos. France Let, 1–36. d KEY RESOURCES TABLE 5. Wu, L.Q., and Dickman, J.D. (2011). Magnetoreception in an avian brain in part mediated by inner ear lagena. Curr. Biol. 21, 418–423. d LEAD CONTACT AND MATERIALS AVAILABILITY d EXPERIMENTAL MODEL AND SUBJECT DETAILS 6. Bellono, N.W., Leitch, D.B., and Julius, D. (2018). Molecular tuning of electroreception in sharks and skates. Nature 558, 122–126. d METHOD DETAILS B Magnetically shielded room and magnetic coils 7. Landler, L., Nimpf, S., Hochstoeger, T., Nordmann, G.C., Papadaki- Anastasopoulou, A., and Keays, D.A. (2018). Comment on B Subjects and Stimulations ‘‘Magnetosensitive neurons mediate geomagnetic orientation in B Immunohistochemistry on brain sections Caenorhabditis elegans.’’ eLife 7, e30187. B Counting and statistical analysis 8. Karten, H.J., and Hodos, W. (1967). A Stereotaxic Atlas of the Brain of the B Generation of cell density maps Pigeon (Columba livia) (The Johns Hopkins Press). B Modeling of electromagnetic induction in an artificial 9. Dickman, J.D., and Fang, Q. (1996). Differential central projections of pigeon semicircular canal vestibular afferents in pigeons. J. Comp. Neurol. 367, 110–131. B Physical calculations 10. Nordmann, G.C., Hochstoeger, T., and Keays, D.A. (2017). B Immunohistochemistry on semicircular canal sections Magnetoreception—a sense without a receptor. PLoS Biol. 15, e2003234.

4058 Current Biology 29, 4052–4059, December 2, 2019 11. Lauwers, M., Pichler, P., Edelman, N.B., Resch, G.P., Ushakova, L., 27. Vincent, P.F., Bouleau, Y., Charpentier, G., Emptoz, A., Saﬁeddine, S.,

Salzer, M.C., Heyers, D., Saunders, M., Shaw, J., and Keays, D.A. Petit, C., and Dulon, D. (2017). Different CaV1.3 channel isoforms control (2013). An iron-rich organelle in the cuticular plate of avian hair cells. distinct components of the synaptic vesicle cycle in auditory inner hair Curr. Biol. 23, 924–929. cells. J. Neurosci. 37, 2960–2975. 12. Nimpf, S., Malkemper, E.P., Lauwers, M., Ushakova, L., Nordmann, G., 28. Kollmar, R., Fak, J., Montgomery, L.G., and Hudspeth, A.J. (1997). Hair Wenninger-Weinzierl, A., Burkard, T.R., Jacob, S., Heuser, T., Resch, cell-specific splicing of mRNA for the alpha1D subunit of voltage-gated G.P., and Keays, D.A. (2017). Subcellular analysis of pigeon hair cells im- Ca2+ channels in the chicken’s cochlea. Proc. Natl. Acad. Sci. USA 94, plicates vesicular trafficking in cuticulosome formation and maintenance. 14889–14893. eLife 6, e29959. 29. Holt, C., Campbell, M., Keays, D.A., Edelman, N., Kapusta, A., Maclary, E., 13. Jandacka, P., Burda, H., and Pistora, J. (2015). Magnetically induced Domyan, E.T., Suh, A., Warren, W.C., Yandell, M., et al. (2018). Improved behaviour of ferritin corpuscles in avian ears: can cuticulosomes function genome assembly and annotation for the rock pigeon (Columba livia). G3 as magnetosomes? J. R. Soc. Interface 12, 20141087. (Bethesda) 8, 1391–1398. 14. Malkemper, E.P., Kagerbauer, D., Ushakova, L., Nimpf, S., Pichler, P., 30. Modrell, M.S., Bemis, W.E., Northcutt, R.G., Davis, M.C., and Baker, C.V. Treiber, C.D., de Jonge, M., Shaw, J., and Keays, D.A. (2019). No evidence (2011). Electrosensory ampullary organs are derived from lateral line plac- for a magnetite-based magnetoreceptor in the lagena of pigeons. Curr. odes in bony fishes. Nat. Commun. 2, 496. Biol. 29, R14–R15. 31. Jørgensen, J.M. (2005). Morphology of electroreceptive sensory organs. 15. Faraday, M. (1832). V. Experimental researches in electricity. Philos. In Electroreception, T.H. Bullock, C.D. Hopkins, A.N. Popper, and R.R. Trans. R. Soc. Lond. 122, 125–162. Fay, eds. (Springer), pp. 47–67. 16. Jungerman, R.L., and Rosenblum, B. (1980). Magnetic induction for the 32. Baker, C.V., Modrell, M.S., and Gillis, J.A. (2013). The evolution and devel- sensing of magnetic fields by animals—an analysis. J. Theor. Biol. 87, opment of vertebrate lateral line electroreceptors. J. Exp. Biol. 216, 2515– 25–32. 2522. 17. Sauer, G., Richter, C.-P., and Klinke, R. (1999). Sodium, potassium, chlo- 33. Trincker, D. (1957). [Structural potentials in the semicircular canal system ride and calcium concentrations measured in pigeon perilymph and endo- of guinea pigs & their changes in experimental deviations of the cupula]. lymph. Hear. Res. 129, 1–6. Pflugers Arch. Gesamte Physiol. Menschen Tiere 264, 351–382. 18. Malkemper, E.P., Mason, M.J., Kagerbauer, D., Nimpf, S., and Keays, D.A. 34. Cohen, M.M. (1981). Intersensory Perception and Sensory Integration (2018). Ectopic otoconial formation in the lagena of the pigeon inner ear. (Plenum Press). Biol. Open 7, bio034462. 35. Keeton, W.T. (1971). Magnets interfere with pigeon homing. Proc. Natl. 19. Feynman, R.P., Leighton, R.B., and Sands, M. (2011). The Feynman Acad. Sci. USA 68, 102–106. Lectures on Physics, Vol. II: The New Millennium Edition: Mainly 36. Ioale` , P. (2000). Pigeon orientation: effects of the application of magnets Electromagnetism and Matter (Basic Books). under overcast skies. Naturwissenschaften 87, 232–235. 20. Kano, F., Walker, J., Sasaki, T., and Biro, D. (2018). Head-mounted sen- 37. Moore, B.R. (1988). Magnetic fields and orientation in homing pigeons: ex- sors reveal visual attention of free-flying homing pigeons. J. Exp. Biol. periments of the late W. T. Keeton. Proc. Natl. Acad. Sci. USA 85, 4907– 221, jeb183475. 4909. 21. Kalmijn, A.J. (1982). Electric and magnetic field detection in elasmobranch 38. Mouritsen, H., Huyvaert, K.P., Frost, B.J., and Anderson, D.J. (2003). fishes. Science 218, 916–918. Waved albatrosses can navigate with strong magnets attached to their 22. Kalmijn, A.J. (1971). The electric sense of sharks and rays. J. Exp. Biol. 55, head. J. Exp. Biol. 206, 4155–4166. 371–383. 39. Bonadonna, F., Bajzak, C., Benhamou, S., Igloi, K., Jouventin, P., Lipp, 23. Bellono, N.W., Leitch, D.B., and Julius, D. (2017). Molecular basis of H.P., and Dell’Omo, G. (2005). Orientation in the wandering albatross: ancestral vertebrate electroreception. Nature 543, 391–396. interfering with magnetic perception does not affect orientation perfor- 24. Baker, C.V.H., and Modrell, M.S. (2018). Insights into electroreceptor mance. Proc. Biol. Sci. 272, 489–495. development and evolution from molecular comparisons with hair cells. 40. Kirschvink, J.L. (1992). Uniform magnetic fields and double-wrapped coil Integr. Comp. Biol. 58, 329–340. systems: improved techniques for the design of bioelectromagnetic ex- 25. Hafidi, A., Beurg, M., and Dulon, D. (2005). Localization and develop- periments. Bioelectromagnetics 13, 401–411. mental expression of BK channels in mammalian cochlear hair cells. 41. R Development Core Team (2013). R: a language and environment for sta- Neuroscience 130, 475–484. tistical computing (R Foundation for Statistical Computing). 26. Roux, I., Safieddine, S., Nouvian, R., Grati, M., Simmler, M.C., Bahloul, A., 42. Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Perfettini, I., Le Gall, M., Rostaing, P., Hamard, G., et al. (2006). Otoferlin, Pietzsch, T., Preibisch, S., Rueden, C., Saalfeld, S., Schmid, B., et al. defective in a human deafness form, is essential for exocytosis at the audi- (2012). Fiji: an open-source platform for biological-image analysis. Nat. tory ribbon synapse. Cell 127, 277–289. Methods 9, 676–682.

Current Biology 29, 4052–4059, December 2, 2019 4059 STAR+METHODS

KEY RESOURCES TABLE

REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Rabbit anti C-FOS Santa Cruz Cat#sc-25; RRID: AB_2231996 Rabbit anti CACNA1D Alomone labs Cat#ACC-005 Mouse anti MaxiKalpha Santa Cruz Cat#sc-374142; RRID: AB_10919141 Mouse anti acetylated Tubulin Sigma Aldrich Cat#T6793; RRID: AB_477585 Goat anti Otoferlin Santa Cruz Cat#sc-50159; RRID: AB_785002 Mouse anti Syntaxin-1 (SP6) Santa Cruz Cat#sc-20036; RRID: AB_628316 Donkey anti rabbit Alexa Fluor-488 Thermo Fischer Cat#A-21206; RRID: AB_2535792 Donkey anti goat Alexa Fluor-568 Thermo Fischer Cat#A-11057; RRID: AB_2534104 Donkey anti mouse Alexa Fluor-647 Thermo Fischer Cat#A-31571; RRID: AB_162542 Chemicals, Peptides, and Recombinant Proteins Paraformaldehyde Sigma Aldrich S9378 Diaminobenzidine Sigma Aldrich D5905 Hydrogen peroxide Merck 822287 D-mannitol Sigma Aldrich M4125 Neg-50 frozen section medium Thermo Fischer 6502 Triton X-100 Sigma Aldrich X100 Donkey serum Abcam ab7475 Fluorescent Mounting Medium Dako S302380 AluI NEB R0137S Critical Commercial Assays Antigen Unmasking Solution Vector Laboratories H-3301 VECTASTAIN Elite ABC HRP Kit Vector Laboratories PK-6100 RNeasy mini kit Quiagen 74104 Quantitect reverse transcription kit Quiagen 205313 Phusion Hot Start Flex NA polymerase NEB M0535S TOPO-TA Cloning Kit Thermo Fischer 45-0030 Oligonucleotides CACNA1D_27F: CAGGAGTGTTCACTGTTGA This paper N/A CACNA1D_29R: TATTGGCAGCATAGTAGACGT This paper N/A KCNMA1_8F: CAGCCACTAACGTATTGG This paper N/A KCNMA1_10R: TCGCTACGTGCCAG This paper N/A Software and Algorithms Pannoramic Viewer, Version 1.15.4 3D Histech N/A Deﬁniens Architect XD Deﬁniens Software N/A R[40] N/A iTOL v4 https://itol.embl.de/ N/A Prism software for graphs and statistical analysis, version 7 GraphPad Software N/A FIJI [8] N/A Other Magnetically shielded room Magnetic shields N/A 3D Helmholtz coil system Serviciencia BH1300-3-A DC power source Aim TTi CPX400DP Slide scanner 3D Histech Pannoramic 250 FlashIII Nanovoltmeter Keithley Nanovoltmeter 2182A (Continued on next page) e1 Current Biology 29, 4052–4059.e1–e4, December 2, 2019 Continued REAGENT or RESOURCE SOURCE IDENTIFIER Cryostat Thermo Fischer MICROM HM 560 Laser scanning confocal microscope Zeiss LSM780 Tissue lyser II Quiagen 128091236

LEAD CONTACT AND MATERIALS AVAILABILITY

Further information and requests for reagents may be directed to and will be fulﬁlled by the lead contact, David A. Keays (keays@imp. ac.at). This study did not generate new unique reagents.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Male and female adult rock pigeons (Columba livia) from our Austrian cohort were maintained on a 12:12 light-dark cycle at 25Cina custom-built aviary. For the magnetic stimulation assay, 45 experimental birds were used. Six animals were used for tissue cDNA preparations and 6 used for immunohistochemical experiments. Animals were housed and experimental procedures were performed in accordance with an existing ethical framework (GZ: 214635/2015/20) granted by the City of Vienna (Magistratsabteilung 58).

METHOD DETAILS

Magnetically shielded room and magnetic coils The experiments were performed inside a 4.4 m long, 2.9 m wide and 2.3 m high room which was shielded against oscillating electromagnetic fields by a 5 mm thick aluminum layer and against static magnetic fields by a 1 mm thick layer of mu-metal (Figure 1A) (Magnetic Shielding, UK). The ambient magnetic field intensity inside the room was attenuated to 0.3 mT and maximum radio frequency intensities between 0.5 to 5 MHz were below 0.01 nT. Magnetic fields were generated using a double-wrapped, custom-built 3D-Helmholtz coil system (Serviciencia, S. L) situated in the center of the shielded room (Figure 1B). Double wrapped coils permit the generation of magnetic stimuli when current flows in parallel through the coils. In contrast in the control condition current flows antiparallel, generating the same heat and vibration to the experimental situation but no magnetic stimulus [40]. The diameters of the coils were: 1,310 mm (x axis), 1,254 mm (y axis), and 1,200 mm (z axis). The coils were driven by DC power sources (TTI CPX400DP) and a computer situated outside the magnetically shielded room. All cables leading into the room were filtered for radio frequencies.

Subjects and Stimulations Adult pigeons (Columba livia) underwent a surgical procedure to glue (UHU, 37420) a 3D printed polylaurinlactam non-magnetic head-stud to the pigeon skull to restrict head movement during the experiment (Figures 1C and 1D). The animals were habituated to our shielded room, the body harness, and the head fixation apparatus for 30 minutes on 3 consecutive days prior to the experiment. Pigeons were then exposed to a rotating magnetic field of 150 mT (n = 22) or to a zero magnetic field (n = 23). The magnetic field applied replicated that employed by Wu and Dickman. Briefly, it consisted of 360 rotations with 6 steps every 2 s (120 s in total). After each rotation the plane was shifted by 15 and the procedure repeated. This occurred 12 times (a full 360) around the x axis and then for the y- and z-axes. For the zero-magnetic field control the same protocol was used but currents ran antiparallel through the double-wrapped coils, producing a null magnetic field. Exposure to magnetic stimuli lasted for 72 minutes and was performed either in darkness (n = 30) or under white light (400-700 nm, intensity of 760 Lux, n = 15). The experimenters were blind to the stimulation conditions at all times. At the completion of the experiment birds were immediately sacrificed and perfused intracardially with 200 ml of 4% PFA in PBS (Sigma Aldrich, 158127). The brains were dissected, postfixed in 4% PFA for 18 hours at 4C, dehydrated in 30% sucrose (Sigma Aldrich, S9378) for 3 days at 4C and sectioned in the coronal plane on a sledge microtome (40 mm).

Immunohistochemistry on brain sections The vestibular nuclei were identiﬁed using a pigeon brain atlas [8]. Three sections between stereotaxic coordinates P 2.25 and P 2.50 were selected for analysis. Sections were mounted on glass slides, dried for 2 days at room temperature followed by 3 washes in PBS (5 minutes each). Antigen retrieval was then performed in a water bath using antigen unmasking solution heated up to 90C over 1 hour (Vector Laboratories, H-3301). After another washing step (3x5 minutes in PBS), slides were incubated with the C-FOS antibody (1:1500, Santa Cruz, sc-253) in 4% milk/0.3% Triton X-100/PBS for 16 hours at room temperature. Slides were washed 3x5 minutes in PBS, incubated with the secondary antibody (1:1000, anti-rabbit, Vectastain Elite ABC HRP Kit, PK-6100, Vector Lab- oratories) for 2 hours at room temperature, followed by another washing step and incubation with the AB reagent (Vectastain Elite ABC HRP Kit, PK-6100, Vector Laboratories) for 1 hour. After another round of washing, slides were incubated in 0.06% Diamino- benzidine (Sigma Aldrich, D5905) in PBS supplemented with 0.08% H2O2 (Merck, 822287) for 1 minute, followed by 3x5 minute washes in PBS, dehydration in serial dilutions of ethanol and coverslipping. All sections of all birds underwent the staining procedure at the same time to minimize variation in background staining.

Current Biology 29, 4052–4059.e1–e4, December 2, 2019 e2 Counting and statistical analysis Slides were scanned on a slide scanner with a 20x objective (Pannoramic 250 Flash III, 3DHistech) and the vestibular nuclei manually segmented (6 bilateral segments from 3 sections) using Pannoramic Viewer (Pannoramic Viewer 1.15.4, 3DHistech). Segments were exported as TIFF files for further analysis. Automated identification and counting of C-FOS positive nuclei was performed by custom made rule-sets using a machine-learning algorithm embedded in the Definiens Architect software (Definiens Architect XD, Definiens Software). The number of C-FOS positive cells per mm2 was calculated. To analyze the effects of the magnetic stimulation on neuronal activation we performed two-way ANOVAs, using the factors: magnetic treatment, light condition and their interaction. Prior to execution of this statistical test we checked all four groups using the Shapiro-Wilk normality test and found that all groups did not differ from normality. We used the software R [41] for all statistical analyses. Figures were generated using Graphpad Prism (Prism 7 for Mac OS X).

Generation of cell density maps The mean distance betweenP C-FOS positive nuclei was used to generate heatmaps. Speciﬁcally, we employed the following algorithm: I = 1/O(2*MDo+ MD)/(2+N), where I in the intensity, MDo is the mean distance to boarder of a neighboring positive cell, MD is the sum of the mean distances, and N is the number of neighboring cells. Heatmaps were averaged for each treatment group by registration onto a reference template. Section borders were corrected for mapping errors. From averaged heatmaps, a differential C-FOS positive cell density map was generated comparing control and magnetic conditions. All image-processing steps were performed using custom-written macros in Fiji [42].

Modeling of electromagnetic induction in an artificial pigeon semicircular canal Polyurethane tubing (0.8 cm inner diameter, 1.2 cm outer diameter) was used to build a replica of a pigeon semicircular canal with a diameter of 21 cm and a circumference of 69 cm (Figure 3A). The tubing was filled with artificial pigeon endolymph consisting of

141.35 mM potassium, 0.23 mM calcium and 141.81 mM chloride in monoQ H2O[17]. The osmolarity was adjusted to 293 mOsm/L with D-mannitol (Sigma Aldrich, M4125) and pH was adjusted to 7.4. Both ends of the loop were closed with gold plated electrodes and connected to a nanovoltmeter (Keithley Nanovoltmeter Model 2182A). The tubing was positioned in the center of our Helmholtz coil system and exposed to a rotating magnetic field of 150 mT around the vertical axis (6 step changes every 2 s, 120 s in total). The induced voltage in the artificial semicircular canal was measured using the nanovoltmeter. The voltage measurement was performed every 10 ms with signal integration between two measurement points to enhance the signal-to-noise ratio. Af- terward, the recorded signal was filtered with a digital Butterworth filter (library functions butter and filtfilt part of the python package signal in scipy; the used parameters were as follow: order of the filter = 5, critical frequency = 0.15) to reduce the low frequency background from the measurement system.

Physical calculations To estimate the diameter of each semicircular canal we measured from the center of the bony canal to the vestibule (n = 3 birds), drawing on previously published CT reconstructions of the pigeon inner ear [18]. The induced voltage, u,is defined as u = ðvFB =vtÞ, where FB is the magnetic flux. Using the geometry of the set-up this expression can be written as u = ðvB =vtÞ,A, with B the applied magnetic field and A the area enclosed by a semicircular canal. The electric field, E, in the semicircular canal is calculated with E = ðu =2prÞ, where r is the radius of the semicircular canal and u the measured induced voltage. We then applied a scaling factor which was calculated by dividing the radius of one semicircular canal by the radius of the artificial inner ear. To estimate the electric field, En, generated by natural head movement we employed the following equation: En = B0, pf, r, where r is the radius of the semicircular canal, B0 is the Earth’s field, and f is the frequency of head scanning.

Immunohistochemistry on semicircular canal sections Pigeon temporal bones containing intact inner ears were ﬁxed overnight in 4% PFA. Following a wash in PBS, ampullae in the mem- branous labyrinth were dissected under a stereomicroscope. Tissues were dehydrated in 30% sucrose/PBS overnight, embedded in Neg-50 Frozen Section Medium (Thermo Fisher, 6502) and sectioned (10 mm) on a cryostat (Thermo Fisher, MICROM HM 560). Slides were dried at RT for at least 1 h prior to staining, washed in 3 3 5 min in PBS, and exposed to heat-based antigen retrieval (Vector Laboratories, H-3301). After a 30 min cooling period and 3 3 5 min washes in PBS, slides were incubated overnight with antibodies diluted in 0.3% Triton X-100/PBS (Sigma Aldrich, X100) with 2% donkey serum (Abcam, ab7475). The following concentrations were used: 1:300 Cav1.3 (Alomone labs, ACC-005), 1:300 BK (Santa Cruz, sc-374142), 1:500 acetylated Tubulin (Sigma-Aldrich, T6793), 1:300 Otoferlin (Santa Cruz, sc-50159), 1:500 Syntaxin 1 (Santa Cruz, sc-20036). The next day, slides were washed 3 x in PBS for 5 min and incubated with donkey anti-rabbit Alexa Fluor-488 (Thermo Fisher, A-21206), donkey anti-goat Alexa Fluor-568 (Thermo Fisher, A-21206), and donkey anti-mouse Alexa Fluor-647 (Thermo Fisher, A-31571), each diluted 1:500, for 1 h at 4cC. After coun- terstaining with DAPI and 3 3 5 min washes in PBS, slides were mounted with Fluorescent Mounting Medium (Dako, S302380). Im- ages were acquired using a laser scanning confocal microscope (Zeiss, LSM780), and processed in ImageJ [42].

PCR analysis of Cav1.3 (CACNA1D) splice isoforms Tissue samples (brain, pineal gland, retina, cochlea, vestibular epithelia, respiratory cochae, skin, heart, muscle, liver, and spleen) were collected from adult pigeons (Columba livia), snap-frozen in liquid nitrogen, and mechanically homogenized with a tissue lyser e3 Current Biology 29, 4052–4059.e1–e4, December 2, 2019 (QIAGEN Tissue Lyser II, 128091236). For the vestibular epithelia the ampullae, lagena, utricle and saccule were pooled. Total RNA was extracted from tissue lysates using the RNeasy mini kit (QIAGEN, 74104) and reverse transcribed with a Quantitect Reverse Tran- scription Kit in accordance with the manufacturers’ instructions (QIAGEN, 205313). The cDNA libraries were diluted to a working concentration of 1:100 and stored at 20C. Polymerase chain reaction (PCR) primers ﬂanking exon 28a were designed based on available genomic resources. The primer sequences were F: CAGGAGTGTTCACTGTTGA; and R: TATTGGCAGCATAGTA GACGT. PCR ampliﬁcation of the tissue cDNA libraries was performed with the Phusion Hot Start Flex DNA polymerase (NEB, M0535S). PCR products were analyzed by agarose gel electrophoresis (4%).

PCR analysis of BK (KCNMA1) splice isoforms To determine the tissue specific expression of different BK (KCNMA1) splice isoforms we designed an assay that relies on PCR amplification of a 239 bp fragment that spans exons 8 to 10 followed by a restriction digest. The primer sequences were F: CAGCCAC TAACGTATTGG; and R: TCGCTACGTGCCAG. We amplified the aforementioned cDNA libraries using Phusion Hot Start Flex DNA polymerase (NEB, M0535S). A restriction digest was then performed using AluI (NEB, R0137S) which discriminates between exon 9a (SI isoform; TTT-GCC-AG^C-TAC) and exon 9b (RA isoform; TTT-GCT-CGC-TAC; no target). PCR products were digested following gel purification for one hour at 37C and analyzed by agarose gel electrophoresis (2%). Exon 9a and 9b isoforms amplicons from vestibular tissue were TA-cloned into pCR4-TOPO (Thermo Fischer, 45-0030) in accordance with the manufacturer’s protocol and their sequence confirmed.

Phylogenetic analysis of Cav1.3 (CACNA1D) and BK (KCNMA1) splice isoforms We explored the taxonomic distribution of the 10 amino acid insertion in the pore-forming alpha1D subunit of Cav1.3 (CACNA1D) and RA isoform in the BK (KCNMA1) channel using a sequence-similarity based strategy. To identify CACNA1D homologs containing the

KKER insertion we performed a BLASTp search with a 50 amino acid long segment of pigeon CaV1.3 (centered around the insertion) against the NCBI non-redundant protein database (NP_990365.1:1264-1313; BLASTP v2.8.1+; limited to the top 10000 target sequences). Only sequences with similarity in 5 out of the 10 amino acid positions were kept. In the case of the BK channel a BLASTp search was performed with the RA isoform ± 20 ﬂanking amino acids against the NCBI non-redundant protein database. We derived species phylogenetic trees using NCBI’s Common Tree and iTOL v4.

QUANTIFICATION AND STATISTICAL ANALYSIS

To analyze the effects of the magnetic stimulation on neuronal activation we performed two-way ANOVAs, using the factors: magnetic treatment, light condition and their interaction. Prior to execution of this statistical test we checked all four groups using the Shapiro-Wilk normality test and found that all groups did not differ from normality. We used the software R [41] for all statistical analyses. One animal was excluded from analysis because of methodological problems (sections detached from the slide and could not be further analyzed). The exact sample sizes, means, and standard deviations can be found in the results section of the paper (one ‘‘n’’ is defined as one animal). We defined a significant result with p < 0.05 throughout the paper. The treatment conditions were ran- domized on each test day and the experimenters where blind to the applied treatments.

DATA AND CODE AVAILABILITY

The rule-sets used to quantify C-FOS positive cells in the current study have not been deposited in a public repository as they are embedded in the Deﬁniens Architect software but the used parameters are available from the corresponding author on request.

Current Biology 29, 4052–4059.e1–e4, December 2, 2019 e4